Structures 23 (2020) 799–811

Contents lists available at ScienceDirect

Structures
journal homepage: www.elsevier.com/locate/structures

Cyclic response sensitivity of energy dissipating steel plate fuses T

⁎
Ahmad Siar Mahmood Shah, Saber Moradi
Department of Civil Engineering, Ryerson University, 350 Victoria Street, Toronto, Ontario M5B 2K3, Canada

A R T I C LE I N FO A B S T R A C T

Keywords: Steel plate devices are commonly used for the seismic energy dissipation and damage mitigation of structures.
Steel plate fuse Particularly, butterfly-shaped steel plate fuses are essential components in controlled rocking frames, which have
Energy dissipating device been shown as promising self-centering systems for minimizing financial losses due to repair costs and downtime
Sensitivity analysis in strong earthquakes. The cyclic behavior of a steel plate fuse varies depending on design factors related to its
Statistical design of experiment
geometry and material. This paper aims at assessing the effect of seven factors on the cyclic response of steel
Finite element simulation
plate fuses. These factors include the fuse length, mid-width, end-width, thickness, band zone length, fuse yield
strength, and modulus of elasticity. Detailed finite element models are first developed and validated against past
experimental test results. A design of experiment method is used to statistically evaluate the effects of the design
factors and their interactions on the cyclic response characteristics of steel plate fuses in terms of initial stiffness,
yield strength, ultimate stiffness, effective damping, maximum strength, and ductility. The results show that the
most significant effects are the fuse end-width, thickness, and length.

1. Introduction
The conventional approach to seismic design is based on ductile
behavior of structures. Appropriate strength and inherent ductility of
structures allow inelastic drift capacity and energy dissipation during
severe seismic motions. Steel plates with engineered cuts can be im-
plemented in the structures as the sacrificial ductile structural elements
and easily replaceable fuses. The replaceability of these plates are re-
ferred to as structural fuses, which experience damage and thus limit
the seismic forces applied to the structure [1,2]. By confining the in-
elastic deformation and structural damage to steel plates with shear
links, the primary structural elements like beams and columns remain
essentially elastic and undamaged. One example of using structural
steel plate fuses is in controlled rocking steel braced frames, which have
been shown as promising self-centering systems that can minimize fi-
nancial losses due to repair costs and downtime in structures under
strong earthquakes [3].
Extensive research has been conducted on different configurations
of steel plates to investigate their ductility and energy dissipation
capability. Different shapes and dimensions of steel plate fuses result in
distinct cyclic response characteristics, such as shear deformation and
ductility. The main goal in all varied configurations is to achieve a
higher ductility considering the simplicity and cost-effectiveness of
steel plates [4]. Fig. 1(a) illustrates an example of a steel plate with
straight cut openings or slits, which shows higher ductility among all
previously used solid steel plate fuses [4]. The openings divide the plate
into several parallel links, thereby making the energy dissipating plate
more flexible. As a result, its behavior is mainly dominated by flexural
mode behavior compared to predominant shear mode behavior of solid
steel plates without openings [4].
Previous research has been conducted to improve shear deformation
capacity through improving the buckling resistance of the shear plates
by means of controlling compactness (the width-to-thickness ratio) of
each single link compared to the original solid plate [4] (Fig. 1(b)).
Relatively, steel slit fuses, which mostly possess a link width-to-thick-
ness ratio between 1 and 2, exhibit stable hysteretic behavior up to a
peak shear deformation between 10% and 20% [5,6]. Along the same
lines, supplying additional shear deformation capacity for structural
systems is relied on the out-of-plane buckling of the damper [7]. In a
study by Chan et al. [6], steel plate fuses fractured after reaching their
peak shear deformation at an ultimate strength level twice the amount
of their yield strength [6].
The alteration of the straight openings with varying mid-width
honeycomb-shaped openings, leaving butterfly-shaped links, affects the
yielding capacity, shear deformation, strength, and stiffness of the fuse
[8,9]. The resemblance of butterfly-shaped links (Fig. 2) to the moment
diagram of beams deforming in double curvature results in a wider
yielding distribution along the length of the plate [4,9]. Along the same
lines, in the experimental tests by Kobori et al. [9], butterfly-shaped
fuses exhibited 30% shear deformation without any stiffness

⁎
Corresponding author.
E-mail addresses: ahmad.mahmood@ryerson.ca (A. Siar Mahmood Shah), s.moradi@ryerson.ca (S. Moradi).

https://doi.org/10.1016/j.istruc.2019.12.026
Received 18 September 2019; Received in revised form 3 December 2019; Accepted 23 December 2019
2352-0124/ © 2019 Institution of Structural Engineers. Published by Elsevier Ltd. All rights reserved.
A. Siar Mahmood Shah and S. Moradi
Fig. 1. (a) Steel plate with slits or straight cut openings and (b) details of a single link.
Fig. 2. Details of a fuse link.
degradation [9]. Butterfly-shaped fuses can provide substantial stiffness
and energy dissipation [7].
The use of butterfly-shaped fuses has been investigated for several
types of structures against earthquake loading [10–13]. Butterfly-
shaped steel plate fuses have been used in low- to mid-rise structures
[14–16]. It has been found that, with the use of butterfly-shaped fuses,
the structure will shake less when it is imposed to seismic motions [11].
Butterfly-shaped fuses will reduce the response of the structure by
20%–30% without having the fuse destroyed after every single major
event due to the margin of rotation capability of 0.25–0.5 degrees. An
experimental study [17] reports that butterfly-shaped fuses provide
stable energy dissipation until the fuse is fractured or the strength is
reduced by 25%. In addition to the use of butterfly-shaped fuses, it has
been observed that these fuses can have better alignment of bending
capacity with added damper and stiffness (ADAS) devices [18]. More
recently, butterfly-shaped fuses have found applications in rocking
frames (e.g., [3]).
The performance of bridge systems with butterfly-shaped fuses also
has been investigated in previous research (e.g., [19–21]). The im-
plementation of butterfly-shaped fuses increased the stiffness, strength,
and energy dissipation of the bridge system by as high as 40% when
compared to a bare system [21]. Shear links with shorter lengths were
found to have uniformly distributed inelastic shear strains, thus al-
lowing large inelastic deformations without having significant local
strains. In another study [21], long links were shown to experience
large inelastic deformations with significant bending stresses at the
ends of the fuse link.
Lee et al. [22,23] have experimentally tested three shapes of non-
uniform (i.e., varying thickness) and uniform strip dampers to have a
deeper understanding of the behavior of steel plate fuses. Straight cut
800
Structures 23 (2020) 799–811
dampers also have been studied in parallel to butterfly-shaped links
with uniform and non-uniform thicknesses. The experimental results
have shown that the butterfly-shaped links provide a better hysteretic
behavior under cyclic loading when compared to straight cut dampers.
The geometry of straight damper concentrates the damage to the fuse
link ends, thus ultimately, decreasing the amount of energy being dis-
sipated in the system when compared to butterfly-shaped fuses [22].
For improved vibration control of the structure, the use of high ductile
straight and butterfly-shaped fuses with a lower yielding point steel has
been recommended. In the experimental tests, 10 mm to 17 mm thick
steel plates have been used to provide enough resistance against
buckling [22].
Several studies have investigated the effect of taper ratio (mid-width
to end-width ratio, a/ b , as shown in Fig. 2) on the plastic hinge for-
mation, buckling, and energy dissipation of the fuse considering plastic
strains of fuses with various thicknesses [24,25]. As the butterfly-
shaped links get straighter, the location of plastic hinges approaches the
ends of the shear link. However, for the butterfly-shaped links with a
smaller mid-width, the geometrical hardening response starts in lower
displacements right after the first occurrence of yielding [26]. A recent
study [24] reports that plastic strain values are five times higher for the
links with equal end-width and mid-width values compared to but-
terfly-shaped links, which explains the possibility of early fracture in
straight cut steel plate fuses. Past research [26] has shown the sig-
nificant effect of plate thickness on the buckling behavior and energy
dissipation of fuses. Since buckling will cause the plate to fail at an early
stage, it is favorable to design the fuse in a manner that shear or flexural
yielding modes dominate [25]. A plate thickness of more than 0.5 in.
(or 12.7 mm) has been recommended to achieve sufficient energy dis-
sipation [25]. Considering plastic strain results, Farzampour and Ea-
therton [24] have recommended using a/ b ratios of 1/3 and 3/4 for
lower possibility of fracture in the fuse.
Research on the response characteristics of energy dissipating fuses
generally can result in modifications or improvements to the seismic
design requirements, such as width to thickness ratio and lateral bra-
cing requirements. For instance, the stiffness and strength level pro-
vided by the fuse dictates the expected force and deformation demands
on deformation-controlled elements, which are intended to remain
elastic during the seismic event. Further, seismic design of structures
relies on the adequate energy dissipation of ductile elements. For im-
proved seismic design, the accurate prediction of the fuse response
characteristics is essential. This highlights the importance of research
on how the fuse response characteristics (including energy dissipation
capacity and ductility) will change with respect to different design
factors.
A. Siar Mahmood Shah and S. Moradi
2. Motivation and objective of this study
In this paper, the sensitivity of the cyclic response of steel plate fuses
to different design parameters is assessed. In previous studies (e.g.,
[4,27]) the variability of the fuse cyclic response to a number of in-
dividual parameters and their ratios such as thickness, end-width to
thickness, length to thickness, mid-width to end-width of steel plate
fuses has been addressed. However, there is no previous statistical study
of main factor and interaction effects on the fuse cyclic response. In the
present paper, a statistical design of experiment method is used to ex-
amine the effect of each parameter by changing different design para-
meters together. Furthermore, possible interactions between design
parameters are determined. The presence of an interaction between two
factors is statistically indicated when the response sensitivity for one
factor varies at different levels of another factor.
In this study, first, a detailed finite element model is generated to
accurately capture the response of steel plate specimens against past
experimental and numerical studies [22,28]. Following that, a sensi-
tivity study is performed using a two-level factorial design with seven
factors that potentially influence the fuse cyclic response character-
istics. The selected factors include design factors related to both the fuse
geometry and material. The geometry related factors include the fuse
link length, mid-width, end-width, thickness, and band zone length.
The effect of the fuse yield strength and modulus of elasticity as ma-
terial-related factors are also evaluated. Thirty-two finite element
models of fuses with different design factor combinations are analyzed;
and the cyclic force-displacement response of each steel plate fuse is
assessed in terms of initial stiffness, yield strength, ultimate stiffness,
effective damping, maximum strength and ductility.
3. Finite element model
To assess the cyclic response sensitivity of steel plate fuses, 3D finite
element models are developed and analyzed in the finite element
software, ANSYS Mechanical APDL [29]. Two specimens from the
uniform strip steel dampers tested by Lee et al. [22] are chosen for the
calibration and verification of finite element modeling results under full
cyclic loading. Further, three different single-link specimens from the
numerical study by Farzampour and Eatherton [28,32] are modeled to
verify the accuracy of the finite element analysis results in this study,
which is based on the response of a single link of the fuse. These spe-
cimens exhibit three different limit states, including flexural yielding,
shear yielding, and lateral torsional buckling.
The experimentally tested specimens by Lee et al. [22] include a
conventional prismatic strip damper (PSD, which has slit or straight
openings) and a dumbbell-shaped damper (DSD also known as
Fig. 3. Details of specimen PSD-5 used for the finite element analysis verification (all dimensions are in mm) [22].
801
Structures 23 (2020) 799–811
butterfly-shaped fuses), as shown in Figs. 3 and 4, respectively. Table 1
summarizes the details and dimensions of steel plate fuses used for
verification purposes. In Table 1, r denotes the radius of the cut at the
end of the link and n is the number of the links in the specimen.
For a satisfactory level of accuracy in the finite element simulations
of highly nonlinear steel plate fuses, several trial models were analyzed.
To reduce computational cost (i.e. running time duration and demand
for resources), only one link of the fuse specimen is modeled excluding
experimental setup components. Proper boundary conditions are di-
rectly applied to the fuse model to simulate the accurate behavior of the
fuse. Initial trial runs were also explored with different material prop-
erties and element types.
In the test setup by Lee et al. [22], the bottom of the strip damper is
sandwiched between two thick angles that are bolted to a rigid reaction
beam. This is simulated in the model by fixing all the degrees of
freedom at the bottom of the plate at the bolt line. At the very top of the
test setup, the loading beam is constrained against out-of-plane trans-
lation at all locations and it is supported by two vertical struts on two
sides of the loading beam. The loading beam is bolted to the top edge of
the plate where it is constrained against out-of-plane as well as vertical
translation. This is simulated in the finite element model by assuming a
roller reaction at the top edge of the specimen where it can only move
in the direction of loading and is constrained against out-of-plane and
vertical translation. In the test setup, the load is applied at the top edge
of the loading beam whereas in the finite element models, the load is
applied at the top edge of the plate. It is also worth mentioning that the
band zones at the top and bottom of the plate are modeled to better
capture the stiffness of the model. The finite element model is shown in
Fig. 5.
SHELL181 elements with reduced integration are used for the
models in this study. This element type is suitable for analyzing thin to
moderately thick shell structures. SHELL181 is a four-node element
with six degrees of freedom at each node: 3 translational (Ux, Uy, Uz)
and 3 rotational (ROTX, ROTY, ROTZ) degrees of freedom. From the
ANSYS material library, several types of material models are compared
to check the behavior of the model. The hysteretic curves are then
compared with the experimental results to find the best matching ma-
terial model for the numerical simulations. Different material models
were used, including bilinear elastic perfectly plastic, multilinear ki-
nematic hardening model, the NLISO model (Chaboche model com-
bined with nonlinear isotropic hardening) and combined isotropic and
kinematic hardening model. As a result of these comparisons, it was
concluded to select NLISO model (Chaboche model combined with non-
linear isotropic hardening) as the material model. An elastic modulus of
200 GPa and yield stress of 295 MPa are used for steel based on the
coupon tests reported in Lee et al. [22].
A. Siar Mahmood Shah and S. Moradi Structures 23 (2020) 799–811

Fig. 4. Details of specimen DSD-5 used for verification purposes (all dimensions are in mm) [22].

Table 1 important to create fine enough and regular meshing in the model for
Detailed dimensions of the specimens used for verification purposes [22,28]. capturing accurate cyclic response and overcoming convergence pro-
Specimen L (mm) a (mm) b (mm) t (mm) r (mm) n blems.
An element size of 3.5 mm is used for the fuse model based on a
PSD-5 180 36 36 10 5 10 mesh sensitivity study. This mesh size will result in having 6, 20 and 46
DSD-5 180 18 36 10 5 10 elements over the mid-width, end-width and length, respectively, fol-
Specimen with shear yielding 1000 100 500 20 5 1
lowing the recommendations by Farzampour and Eatherton [28,32].
Specimen with flexural 1000 100 300 20 5 1
yielding The finite element model for specimen DSD-5 for instance, consists of
Specimen with lateral torsional 500 100 100 16.67 5 1 108 key-points, 152 lines, 40 areas, 6662 nodes, 5944 elements (Fig. 6).
buckling It is confirmed that imperfection is not significantly influencing the
response of steel fuse models. Therefore, no imperfection is introduced
in the models.
The NLISO material model needs the specification of three different Displacement-controlled loading is used in the analyses. Similar to
constants, including; R0 , R∞, and b0 . These values characterize the the reference experimental study [22], the loading protocol consists of
isotropic hardening behavior of the material. R0 and R∞ are the initials symmetrical cyclic drift ratio with incrementally increasing amplitudes.
isotropic hardening modulus, R0 is the asymptotic hardening, and b0 The amplitude of each single step and the number of cycles for each
controls the decrease of hardening in the function of equivalent plastic step allow the specimen to undergo a large range of cumulative plastic
strain. For the verified models, values of 16, 30 and 50 are used for R0 , deformation [22]. The amplitudes of the loading for the first 12 steps is
R∞, and b0 , respectively [30]. increased with a ratio of 1.40. For instance, the 1st load step is set to
After defining the element type and material properties, mesh 0.44 mm and the 2nd load step is obtained by the multiplication of 1.40
control features in ANSYS Mechanical APDL [29] are used to control and 0.44 (i.e., the second load step = 1.40 × 0.44 mm = 0.62 mm).
the mesh density. These features include assigning line and element Following the 12th step, the increments have been fixed at 5.40 mm.
sizes. An automated free mesh, which gives the most desirable regular Fig. 7 shows the loading program.
meshing in irregular shapes is generated for the entire model. Fig. 6 Nonlinear static analyses with full Newton-Raphson solution
shows the meshing for the finite element model of specimen DSD-5. It is scheme are performed in ANSYS [29]. Geometric and material

Fig. 5. Finite element model for specimen PSD-5.

802
A. Siar Mahmood Shah and S. Moradi
Fig. 6. Meshing of specimen DSD-5.
nonlinearities are included in the displacement-control analysis. In
order to avoid convergence difficulties, the minimum number of sub-
steps to be taken is set to 10 and 25 for small and large amplitude load
steps, respectively.
3.1. Validation of finite element analysis
This section presents the validation of the finite element modeling
for different specimens, including two steel plate fuses (DSD-5 and PSD-
5) among those specimens experimentally tested by Lee et al. [22] and
three single-link specimen models each with different governing limit
states [28,32]. Fig. 8 shows the lateral load versus lateral displacement
of specimens DSD-5 and PSD-5 in comparison with the experimental
response curves. As shown in Fig. 8, the numerical results are in good
agreement with the cyclic test results. The error in predicting the peak
strength using the finite element model is 1.14% and 3.50% for speci-
mens PSD-5 and DSD-5, respectively. The error percentage here is ob-
FE − PTest
Pmax max FE Test
tained from Test × 100 , where Pmax and Pmax are the peak strength
Pmax
from the finite element analysis and the experimental test, respectively.
The accuracy of the finite element analysis in predicting the limit
states of butterfly-shaped fuses is also verified against three specimen
models analyzed by Farzampour [28,32]. The detailed dimensions of
the models experiencing flexural yielding, shear yielding, and lateral
torsional buckling are listed in Table 1. These models include only one
link of the butterfly-shaped and slit fuses to reduce the computational
cost. As the presented study is also considering one link of the fuse, the
finite element model is validated with the reference numerical results.
Fig. 7. Cyclic loading profile.
803
Structures 23 (2020) 799–811
Note that the reference study is also based on validations against ex-
periments [26]. The lateral load-displacement plots of these three
specimens are compared and shown in Fig. 9.
3.2. Sensitivity analysis
A sensitivity analysis of the cyclic load-displacement response of
steel plate fuses is conducted to assess the effects of different varying
design parameters based on the statistical design of experiment method
and the validated finite element models. Seven potentially influential
factors on the cyclic response characteristics are considered. Five of the
factors are related to the fuse geometry, including the fuse length, L ;
mid-width, a ; end-width, b ; thickness, t ; and band zone length, c . The
remaining two factors are material-related factors, including steel yield
strength, Fy and modulus of elasticity, E . Two levels are considered for
the input factors. Table 2 lists these factors along with their low (−)
and high levels (+). The factor ranges are selected to be broad and
practical based on previous experimental studies (e.g., [4,22]).
3.3. Design of experiment
The sensitivity analysis is carried out using the Design-Expert soft-
ware [33]. A two-level fractional factorial design is considered for the
analysis. A full factorial design with seven factors and two levels for
each factor requires analyzing 27 (=128) models, which is computa-
tionally expensive. Therefore, a one-fourth fractional factorial design
with a resolution of four has been employed in this study. By doing this,
A. Siar Mahmood Shah and S. Moradi Structures 23 (2020) 799–811

Fig. 8. Lateral load-displacement response of specimens DSD-5 and PSD-5: numerical results in comparison with experimental results [22].

(a) (b)

(c)
Fig. 9. Validation of the finite element modeling against a numerical study of three limit states [28,32]: (a) flexural yielding; (b) shear yielding; (c) lateral torsional
buckling.

Table 2 factor E is specified by the multiplication of L , a , b and t . Table 3 lists

Factors and ranges considered for the sensitivity analysis. the factor combinations for each run. For each combination of factors, a
Factors Symbol Low level (−) High level (+) Units
finite element model has been created and analyzed under cyclic
loading.
Length L 180 500 mm As the output of each simulation, six different response variables
Mid-width a 15 36 mm have been recorded from the cyclic load-displacement response. These
End-width b 36 200 mm
Thickness t 5 40 mm
response parameters include: (1) the initial stiffness, K ; (2) yield
Steel yield strength Fy 245 340 MPa strength, Py ; (3) ultimate stiffness, K ; (4) effective damping, βeff ; (5)
Modulus of elasticity E 185,000 212,000 MPa maximum strength, P ; and (6) ductility, μ . Table 4 lists the response
Band zone length c 5 75 mm values obtained from the nonlinear analysis of each model.
Effective stiffness (k eff ) is calculated from Eq. (1), where the positive
and negative peak displacements (δ + and δ−) and the corresponding
the number of runs is reduced to 27 − 2 (=32). It is noted that 32 or (25) loads (P + and P −) are used. The effective damping is used to convert the
runs is for a full factorial design with five factors including all possible hysteretic energy of a damper into an equivalent linear viscous system.
combinations of high and low levels. Since seven factors are presented Effective damping (βeff ) is used to define the nonlinear behavior of a
in the design, two generators (=7–2) are needed to assign the low or displacement dependent device [31]. It describes the energy dissipation
high level for the remaining factors. These two generators are; E = (L ) capacity of a fuse and is calculated using Eq. (2), in which Eloop , k eff , δ +
(a ) (b ) (t ) and c = (L ) (a ) (b ) (Fy ). In this case, the high and low levels of and δ− define the hysteresis energy, equivalent stiffness, positive and

804
A. Siar Mahmood Shah and S. Moradi Structures 23 (2020) 799–811

Table 3 Table 4
Factor combination for each model. Response characteristics of each model.
Model L : Length a: b: t : Thickness Fy : Steel E: c: Model Initial Yield Ultimate Effective Maximum Ductility
Mid- End- yield Modulus Band Stiffness Strength Stiffness Damping Strength
width Width strength of zone kN/mm kN kN/mm kN
elasticity length
mm mm mm mm MPa MPa mm 1 572.97 212.0 5.19 2600.0 565 107.03
2 959.38 307.0 2.04 2310.0 663 106.88
1 180 36 200 40 245 185,000 75 3 18.070 30.90 2.02 43,600 202 76.842
2 180 36 200 40 340 185,000 5.0 4 79.333 11.90 0.27 346.00 26.0 86.000
3 180 36 36.0 40 340 212,000 75 5 1.6887 8.680 0.40 10,400 50.7 19.260
4 180 15 200 5.0 245 185,000 5.0 6 0.1851 1.090 0.03 354.00 2.03 3.9728
5 500 15 36.0 40 245 212,000 5.0 7 54.545 24.00 0.37 14,300 131 200.45
6 500 15 36.0 5.0 245 185,000 5.0 8 2.6459 13.60 0.18 55,000 99.0 36.070
7 180 36 36.0 40 245 212,000 5.0 9 3.9444 4.260 0.09 344.00 6.97 11.944
8 500 36 36.0 40 340 185,000 5.0 10 1.8278 2.760 0.03 330.00 4.86 11.818
9 180 15 36.0 5.0 340 212,000 5.0 11 5.8000 3.190 0.11 646.00 6.86 32.727
10 180 15 36.0 5.0 245 212,000 75 12 510.71 143.0 5.19 1420.0 405 102.85
11 180 36 36.0 5.0 245 185,000 5.0 13 134.29 37.60 0.73 550.00 63.0 64.286
12 180 15 200 40 340 212,000 75 14 0.3730 1.750 0.02 962.00 3.80 7.2920
13 180 36 200 5.0 340 212,000 5.0 15 0.2412 1.240 0.05 78.300 1.89 2.5097
14 500 36 36.0 5.0 340 212,000 5.0 16 1.3398 12.30 0.06 11,300 59.5 11.372
15 500 36 36.0 5.0 245 212,000 75 17 55.312 17.70 1.25 34.200 25.2 10.469
16 500 15 36.0 40 340 212,000 75 18 15.000 18.30 0.82 173.00 28.8 10.574
17 180 15 200 5.0 340 185,000 75 19 11.590 27.70 0.64 111.00 38.5 5.3975
18 500 36 200 5.0 245 185,000 5.0 20 11.229 13.70 0.30 29.300 18.1 3.8442
19 500 36 200 5.0 340 185,000 75 21 12.748 21.80 0.67 4280.0 58.5 29.474
20 500 15 200 5.0 340 212,000 5.0 22 66.481 71.80 1.15 3020.0 198 41.667
21 180 15 36.0 40 245 185,000 75 23 706.67 106.0 2.04 1190.0 327 192.00
22 500 15 200 40 245 185,000 75 24 1.9037 4.550 0.08 642.00 6.97 9.7908
23 180 15 200 40 245 212,000 5.0 25 79.344 96.80 0.13 3260.0 283 72.295
24 180 36 36.0 5.0 340 185,000 75 26 136.06 166.0 2.59 13,100 372 63.442
25 500 15 200 40 340 185,000 5.0 27 1.6768 11.00 0.37 17,300 62.0 17.561
26 500 36 200 40 245 212,000 5.0 28 106.13 225.0 1.57 14,800 490 39.056
27 500 36 36.0 40 245 185,000 75 29 0.1488 1.690 0.06 139.00 2.68 2.0617
28 500 36 200 40 340 212,000 75 30 27.705 33.80 0.39 19,600 106 76.721
29 500 15 36.0 5.0 340 185,000 75 31 83.125 26.60 0.38 394.00 46.4 40.133
30 180 15 36.0 40 340 185,000 5.0 32 9.6226 10.20 1.18 4.5300 12.8 2.2547
31 180 36 200 5.0 245 212,000 75
32 500 15 200 5.0 245 212,000 75
design factors on the cyclic response characteristics of fuses.
Important factors and interactions that have significant effects on
negative peak displacements under the load-displacement plot, re- the cyclic response are identified by a half-normal probability plot. This
spectively. method is based on the theory that insignificant factors fall on a straight
( |P +| + |P −|) line and are normally distributed with a near-zero mean value [34].
keff = After generating the half-normal probability plot for each response
(|δ +| + | δ−|) (1)
variable, we obtain the effects of main factors and their interactions.
2 Eloop In order to check the significance of the statistical model for each
βeff = ⎛ ⎞ response variable, statistical analysis of variance (ANOVA) using the
π (k
⎝ ⎠ eff ) ∗ (|δ +| + | δ−|)2 (2)
Design-Expert software is performed [33]. The results from half-normal
Initial stiffness (K ) is calculated as the ratio of yield strength to yield probability plots are also confirmed. ANOVA is a method to statistically
displacement while ultimate stiffness (K ) is calculated from the slope of test the null hypothesis of no treatment effects [34]. A significance level
the last two steps of the response curve. Further, the ductility is ex- of 5% is chosen for this ANOVA study. Hence, the most significant
pressed as the ratio of maximum displacement to yield displacement. factors have p-values less than 0.05. For instance, considering the initial
Lastly, maximum load (P ) is considered as the maximum strength of the stiffness response characteristic, the significance level of the factors is
fuse, while the first occurrence of yielding is taken as the yield strength determined based on their p-values. The p-value for the fuse end-width
(Py ). (b ), thickness (t ), and length (L ) is less than 0.0001, which indicates the
significance of their influence on the initial stiffness response.
3.4. Results of the sensitivity analysis In order to ensure the high accuracy of the statistical model, R-
Squared, adjusted R-Squared, predicted R-Squared, and adequate pre-
The load-displacement cyclic response of each model is recorded. A cision are checked. R-Squared value is used to determine the percentage
wide range of cyclic response is observed. For example, the cyclic re- of the data that can be explained by the model. An R-Squared value of
sponse plots for three models 1, 7, and 12 are illustrated in Fig. 10. at least 0.80 is typically desired for a good model. The adjusted R-
Model 7 with the factor combinations shown in Table 3 has shown Squared and predicted R-Squared are used to define the relation of
greater ductility compared to models 1 and 12. The obtained results for experimental data to their predicted values of the final equation. The
the response characteristics of these three models show that the effec- difference between the adjusted R-Squared and the predicted R-Squared
tive damping in model 7 is respectively 5.50 times and 10 times higher should not be greater than 0.20. Adequate precision measures the signal
than those in models 1 and 12. The maximum strength of model 1 is to noise ratio and a value of greater than 4 is desirable [33]. For ex-
4.30 and 1.40 times greater than that in models 7 and 12, respectively. ample, considering the initial stiffness response, the value of R-Squared,
Lastly, the yield strength response is considerably higher in model 1 the difference of adjusted and predicted R-Squared, and the adequate
when compared to model 7 and 12. The sensitivity analysis results will precision are 0.93, 0.05, and 17.84, respectively.
be subsequently presented along with the discussions of the effect of the Percentage contributions of important factors and interactions for

805
A. Siar Mahmood Shah and S. Moradi
(a)
Fig. 10. Hysteretic response plots for: (a) model 1; (b) model 7; (c) model 12.
each response characteristic are presented in Fig. 11. The positive sign
indicates that the corresponding factor or interaction has a positive
effect.
3.5. Sensitivity of response characteristics
3.5.1. Initial stiffness
In this section, the sensitivity of the initial stiffness response of steel
plate fuses is discussed. As shown in Fig. 11(a), the most important
factors considering the initial stiffness are the end-width (b ), thickness
(t ), length (L ), as well as the interaction between end-width and
thickness (bt ), length and end-width (Lb ), length and thickness (Lt ),
and mid-width and modulus of elasticity (aE ) with the percentage
contributions of 22%, 15%, 15%, 13%, 12%, 9% and 7%, respectively.
The remaining factors shown in Fig. 11(a) contribute less to the initial
stiffness.
Among the significant factors, the end-width (b ), thickness (t ) and
their interaction (bt ) show positive effects (i.e., the initial stiffness re-
sponse is directly proportional to these factors). For example, by in-
creasing the thickness from the low level (5 mm) to the high level
(40 mm), the initial stiffness of the fuse is increased. On the other hand,
the length (L ) and its interaction with the thickness (Lt ), its interaction
with the end-width (Lb ), and the interaction between the mid-width
and modulus of elasticity have negative effects on the initial stiffness
(aE ). This observation is also supported by the results of previous re-
search [4,35] reporting that the initial stiffness of the fuse is directly
proportional to the end-width and thickness; and this response is in-
versely proportional to the length. The formula for calculating the in-
itial stiffness is as follows:
2Ebt 3
K=
3L3
It is observed that, the most significant factors influencing the initial
Structures 23 (2020) 799–811
(b)
(c)
stiffness are all present in the previously derived design formula (Eq.
(3)). The equation further supports the positive effect of end-width (b )
and thickness (t ) as well as the negative effect of length (L ), indicating
that the initial stiffness of the fuse is directly proportional to the end-
width and thickness and it is inversely proportional to the length.
The 3D plot in Fig. 12 shows the effect of interaction (bt ) on the
initial stiffness. It is observed that the effect of the thickness (t ) on the
initial stiffness is significantly increased as the end-width (b ) is in-
creased. As the thickness (t ) is increased (from t = 5 mm to 40 mm) at
the lower level of the end-width (b = 36 mm), there is a slight change
in the initial stiffness of the plate (approximately from 0 to 10 kN/mm).
However, at the high level of the end-width (b = 200 mm), with the
increase in the thickness (from t = 5 mm to t = 40 mm), the initial
stiffness is significantly increased from 0 to approximately 400 kN/mm.
3.5.2. Yield strength
Another response value that is studied from the results of sensitivity
analysis is the yield strength of the steel plates. The yield strength is a
significant response characteristic in the steel plate fuses since the fuse
should have enough yielding capacity to stay undamaged under service
loads [22]. The sensitivity analysis results show that the yield strength
of steel plate fuses is mostly affected by the end-width (b ), thickness (t ),
mid-width (a ), length (L ), as well as the interactions bt and ab having
percentage contributions of 29%, 29%, 5%, 2%, 18%, and 5%, re-
spectively as shown in Fig. 11(b). Among all the main factors, only the
fuse length (L ) has a negative impact on the yield strength. In addition,
it can be observed that over 75% of the variability in the fuse yield
strength is controlled by the end-width (b ), thickness (t ) and their in-
teraction (bt ). The analysis results are additionally confirmed by the
yield strength design formula (Eq. (4)) for slit fuses [4,31].
(3) nb2tFy
Py =
3L (4)
806
A. Siar Mahmood Shah and S. Moradi
(a)
(c)
(e)
Fig. 11. Percentage contribution of important factors and interactions for each response characteristic.
It is perceived that the factors influencing the yield strength are all
present in the previously derived design formula (Eq. (4)). It should be
noted that Eq. (4) is derived only for straight (slit) fuses excluding the
effect of mid-width (a ), which is included in the current sensitivity
study. The equation further confirms the positive effect of end-width (b )
and thickness (t ) as well as the negative effect of length (L ) – showing
that the yield strength of the fuse is directly proportional to the end-
width and thickness and it is inversely proportional to the length.
The 3D plot in Fig. 13 shows the interaction between the end-width
(b ) and the thickness (t ) of the steel plate fuse. It can be observed that
increasing both the end-width (b ) and the thickness (t ) from their low
levels (t = 5 mm, b = 36 mm) to high levels (t = 40 mm, b = 200 mm)
increases the yield strength from 0 to approximately 165 kN.
3.5.3. Ultimate stiffness
The ultimate stiffness of steel plate fuses is mostly affected by end-
807
Structures 23 (2020) 799–811
(b)
(d)
(f)
width (b ), thickness (t ), length (L ), and the interaction between length
and thickness (Lt). This observation is likewise the sensitivity results for
the maximum and yield strength, which were found sensitive to b , t ,
and L . The calculated percentage contributions for the main factors and
interactions indicate that these effects (i.e. the effect of b , t , L , c , Lt , bt ,
Lc , and aE ) account for 24%, 20%, 7%, 5%, 8%, 6%, 4% and 3% of the
total response variability, respectively. As illustrated in Fig. 11(c), end-
width (b ), thickness (t ) band zone (c ), and interaction (bt ) have positive
effects while the remaining factors have negative effects.
To determine the effect of interaction between length (L ) and
thickness (t ) on the variability of the ultimate stiffness response, the 3D
plot in Fig. 14 is illustrated. It can be observed that the ultimate stiff-
ness response increases approximately from 0.40 to 2.25 kN as the
thickness (t ) increases from its low level (t = 5 mm) to high level
(t = 40 mm) and the length (L ) decreases from its high level
(L = 500 mm) to its low level (L = 180 mm).
A. Siar Mahmood Shah and S. Moradi
Fig. 12. The effect of interaction between the thickness and the end-width on
initial stiffness.
Fig. 13. The effect of interaction between the end-width and the thickness on
yield strength.
Fig. 14. The effect of interaction between the length and the thickness on ul-
timate stiffness.
3.5.4. Effective damping
This section describes the sensitivity analysis of the effective
damping response. Factors that have a significant effect on the effective
damping are the thickness (t ), end-width (b ), mid-width (a ), steel yield
strength (Fy ), as well as the two-factor interactions: between end-width
and thickness (bt ), between mid-width and thickness (at ), between
thickness and steel yield strength (tFy ), and between end-width and
808
Structures 23 (2020) 799–811
steel yield strength (bFy ). The remaining factors such as the length (L ),
modulus of elasticity (E ) and the band zone length (c ) have shown no
significant effect on the effective damping response characteristics.
From the significant factors, the thickness (t ), mid-width (a ), steel
yield strength (Fy ), the interactions between mid-width and thickness
(at ), thickness and steel yield strength (tFy ) have positive effects with
the percentage contributions of 29%, 8%, 5%, 7%, and 5%, respectively
as shown in Fig. 11(d). On the other hand, the end-width (b ), the in-
teractions between the end-width and thickness (bt ), and between the
end-width and modulus of elasticity (bE ) show negative effects on the
effective damping with percentage contributions of 12%, 11%, and 4%,
respectively.
These results indicate that the effective damping of steel plate fuses
is mainly controlled by the thickness, end-width, and their interaction
having a combined contribution of more than 50%. This trend can be
attributed to the fact that the energy dissipating capacity of steel plate
fuse is increased significantly with an increase of the thickness because
of the delay in the buckling of the fuses. This is further supported by
Farzampour and Eatherton [26], suggesting that the thickness of the
plate is observed to control the buckling limit state of the fuse, and thus
affecting the energy dissipation capability. Further, the negative effect
of the end-width on the effective damping is expected because in-
creasing the end-width over the length causes the fuse to act more like a
solid plate thus leading to less energy dissipation. These observations
are also consistent with the research by Farzampour and Eatherton
[26]. They reported that increasing the end width over the length of the
butterfly-shaped fuses would make the shear mechanism governing
over the flexure, and thus resulting in a reduction in the energy dis-
sipation [26].
To examine the effect of the interaction between the main factors
end-width (b ) and thickness (t ) on the variability of the effective
damping of steel plate fuses, the 3D plot in Fig. 15 is presented. It can be
observed that at the low level of end-width (b = 36 mm), increasing the
thickness (t ) significantly increases the effective damping response (1 to
215,000). In contrast, at the high level of end-width (b = 200), the
effective damping is increased at a lesser rate (from 1 to 5000) due to
the increase in the thickness (t ). This trend is because of the active
interaction between b and t . In other words, since the interaction is
involved, the amount of change in the response because of varying
thickness (t ) is different at different levels of the end-width (b ).
3.5.5. Maximum strength
The shear load capacity of steel plate fuses is monitored as a re-
sponse characteristic. As illustrated in Fig. 11(e), the important factors
that are influencing the load capacity of the fuses are the thickness (t ),
Fig. 15. The effect of interaction between the end-width and the thickness on
effective damping.
A. Siar Mahmood Shah and S. Moradi Structures 23 (2020) 799–811

end-width (b ), mid-width (a ), length (L ), the interaction between end-

width and thickness (bt ), the interaction between length and thickness
(Lt ), and the interaction between mid-width and thickness (at ) with the
percentage contributions of 43%, 23%, 4%, 3%, 16%, 2%, and 3%,
respectively.
The length (L ) and interaction (Lt ) have negative effects while all
the other significant factors and interactions have positive impacts on
the load capacity. The end-width (b ), thickness (t ), and their interaction
(bt ) account for over 80% contribution. The response sensitivity result
can be explained based on the buckling behavior of steel fuses. Past
experimental studies have confirmed that, because of buckling, thin and
long steel plate fuses reach less strength compared to thick fuses. The
experimental study by Lee et al. [22] reports that the load capacity of
steel plate fuses is significantly decreased when the fuse experiences
buckling under cyclic loading.
The response sensitivity results for the maximum strength are also
supported by the previously developed design formula [4]:
Fig. 17. The effect of interaction between the thickness and the length of the
nb2t Fy fuse link on ductility.
P=
2L (5)
deformations. This fact is also confirmed by the experimental study by
It is confirmed that the factors influencing the maximum strength
Ma et al. [4]. They reported that thick strip dampers (steel plate fuses
are all present in the previously derived design formula (Eq. (5)). This
with straight and butterfly-shaped openings) can undergo large de-
equation however is derived only for straight (slit) fuses excluding the
formations and can have an uncompromised performance since
effect of mid-width (a ), which is included in this sensitivity analysis. Eq.
pinching is delayed in the damper [4]. In addition to thickness, the
(5) further confirms the positive effect of thickness (t ) and end-width
importance of length influencing the deformation capacity is also ob-
(b ) as well as the negative effect of length (L ) – indicating that the
served in the experimental study by Ma et al. in [4] reporting that by
maximum strength of the fuse is directly proportional to the end-width
changing the fuse length, the deformation demand in terms of shear
and thickness, and it is inversely proportional to the fuse length.
deformation can be adjusted to a desired value. As shown in Fig. 11(f),
Among the significant interactions, the interaction between the end-
the ductility response also appears to be affected by interactions (Lt ),
width (b ) and the thickness (t ) is noticeable. The bt interaction is illu-
(ab ), (Lc ), and (LE ) with the percentage contributions of 5%, 3%, 3%,
strated in Fig. 16. It can be observed that the maximum load response
and 3%, respectively.
increases from 0 to 400 kN as the end-width (b ) increases from its low
Depending on the factor combination, a wide range of ductility re-
level (b = 36 mm) to high level (b = 200 mm), when the thickness (t ) is
sponse from 2.06 to 200.46 mm/mm is observed. The 3D plot in Fig. 17,
kept at its high level (t = 40 mm). In contrast, at the low level of
illustrates how the (Lt ) interaction influences the ductility response. As
thickness (t = 5 mm), the increase in the end-width (b ) from its low
the thickness increases, for longer fuses, the ductility is increased from
level (b = 36 mm) to high level (b = 200 mm) only increases the
3 to about 40 mm/mm while for shorter fuses, the ductility is increased
maximum load response from 1 to approximately 20 kN.
from 30 to 110 mm/mm when the other factors are at their mean va-
lues.
3.5.6. Ductility
The ductility of steel plate fuses is mostly affected by the thickness 4. Summary and conclusion
(t ), length (L ), band zone length (c ), and end-width (b ) having per-
centage contributions of 30%, 26%, 9%, and 6%, respectively The effects of seven different design parameters on the cyclic re-
[Fig. 11(f)]. The length and the thickness of steel plate fuses are found sponse of steel plate fuses were examined in this paper. These factors
to control over 50% of the ductility response. This fact can be expected included geometry and material related properties. The geometry re-
because steel plate fuses with higher thickness and lower lengths are lated factors included the length, mid-width, end-width, thickness, and
less prone to pinching and thus these fuses are expected to show large the band zone length of the fuse link. The material related factors in-
cluded the yield strength and modulus of elasticity of the steel plate. A
sensitivity analysis was conducted using a fractional factorial design-of-
experiment method to statistically investigate the potential effects of
the design factors as well as any significant interactions between the
main factors. ANSYS Mechanical APDL was used to develop 3D finite
element models and simulate the cyclic and monotonic response of steel
plate fuses. The models were then validated against past experimental
test results. A two-level, one-fourth, fractional factorial design was used
to generate thirty-two factor combinations, which resulted in a sam-
pling database for the sensitivity study. From the finite element analysis
of the models for these factor combinations, the cyclic load-displace-
ment response of steel plate fuses was evaluated in terms of initial
stiffness, effective damping, ductility, load capacity, and yield strength.
Based on the results, the following conclusions can be made:

• The fuse link end-width and thickness significantly influence all the
response characteristics. The fuse length and mid-width are influ-
Fig. 16. The effect of interaction between the end-width and thickness on the ential on some of the response variables.
maximum strength. • By varying modulus of elasticity (from 185,000 MPa to

809
A. Siar Mahmood Shah and S. Moradi
212,000 MPa), steel yield strength (from 245 MPa to 340 MPa), and
the band zone length (from 5 mm to 75 mm), it is observed that the
cyclic performance of steel plate fuses is not significantly affected by
the variation in these factors. The sensitivity analysis shows that the
variability of modulus of elasticity, steel yield strength, band zone in
the response quantities is generally below 1% except for the effec-
tive damping and ductility having percentage contributions close to
or less than 5%.
• The initial stiffness of a steel plate fuse is mostly influenced by the
end-width, thickness, length and their interactions. These para-
meters account for approximately 92% of the total variability in the
initial stiffness response. Further, the length and its interactions
with the thickness and end-width have a negative effect on the in-
itial stiffness of steel plate fuses.
• The yield strength of steel plate fuses is mostly influenced by the
fuse link end-width, thickness, mid-width, and length. These factors
account for about 87% of the total variability in the response. The
effect of end-width and thickness and their interaction is noticeable
among all these factors having a combined percentage contribution
of approximately 75%. Further, except for the fuse length all other
factors have a positive effect on the yield strength response of steel
plate fuses.
• The three important factors that significantly impact the ultimate
stiffness response are the end-width, thickness, and length. These
factors with their interactions account for about 65% of the total
response variability. End-width, thickness and their interaction have
positive effects while the length and its interaction with the plate
thickness have negative effects on the ultimate stiffness response.
• The effective damping of a steel plate fuse is mainly sensitive to the
thickness and the end-width having a combined contribution of
more than 50%. A higher effective damping is observed for thicker
steel plate fuses with a smaller end-width. The effective damping is
increased by about 5 times for thick fuses with a smaller end-width.
• The factors that mostly influence the load capacity of steel plate
fuses are the thickness, end-width, mid-width, and length. These
factors with their interactions account for nearly 94% of the total
variability in the response. Among these, over 80% of the load ca-
pacity variability is controlled by the thickness, end-width and their
interactions. Further, the length and its interaction with thickness
shows a negative effect on the load capacity response while all the
other factors have positive effects. In addition, with increasing the
end-width for thick fuses, the load capacity is increased sig-
nificantly.
•
The most important factors and interactions influencing the ductility
of steel plate fuses are the thickness, length, end-width, the band
zone length, as well as some interactions that have less than 5%
contributions. Overall, these factors and interactions account for
about 85% of the total variability in the response. Among these
important factors, the length and thickness are found to control over
50% of the ductility response variability. Further, it was observed
that shorter and thicker fuses possess greater ductility.
This study and further complementary research contribute to the
efficient, robust, and reliable design of energy dissipating steel plate
fuses, which are key elements in high-performance earthquake resistant
structures. The results from this study can be used to develop predictive
equations for the response of steel plate fuses. In addition, with the
identified significant factors, optimization studies can be performed to
determine the optimal design of energy dissipating steel plate fuses.
Declaration of Competing Interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influ-
ence the work reported in this paper.
810
Structures 23 (2020) 799–811
Acknowledgment
The financial support from the Natural Sciences and Engineering
Research Council of Canada (NSERC) through Discovery Grant and the
Ryerson University Faculty of Engineering and Architectural Science is
gratefully acknowledged.
References
[1] Enrique Martı́nez-Rueda Juan. On the evolution of energy dissipation devices for
seismic design. Earthq Spectra 2002;18:309–46.
[2] Vargas R, Bruneau M. Experimental response of buildings designed with metallic
structural fuses. II. J Struct Eng 2009. https://doi.org/10.1061/(ASCE)0733-
9445(2009) 135:4(394).
[3] Deierlein G, Krawinkler H, Ma X, Eatherton M, Hajjar J, Takeuchi T, et al.
Earthquake resilient steel braced frames with controlled rocking and energy dis-
sipating fuses. Steel Constr 2011;4:171–5. https://doi.org/10.1002/stco.
201110023.
[4] Ma X, Borchers E, Pena A, Krawinkler H, Billington S, Deierlein GG. Design and
behavior of steel shear plates with openings as energy-dissipating fuses. Intern
Report, John A Blume Earthq Eng Center, Stanford Univ 2011.
[5] Nakamura HO, Sugiyama Y, Wada S, Baba MA, Miyabara T. Passive control system
using steel slit damper, part 1: test of steel slit damper. Kanto, Japan: Summ. Tech.
Pap. Annu. Meet. AIJ; 1997. p. 843–4.
[6] Chan RWK, Albermani F, Williams MS. Evaluation of yielding shear panel device for
passive energy dissipation. J Constr Steel Res 2009;65:260–8. https://doi.org/10.
1016/j.jcsr.2008.03.017.
[7] Li G, Li HN. Experimental study and application in steel structure of “dual func-
tions” metallic damper. Adv Steel Constr 2013;9:247–58.
[8] Katayama T, Ito S, Kamura H, Ueki T, Okamoto H. Experimental study on hysteretic
damper with low yield strength steel under dynamic loading, 2000, p. 1–8.
[9] Kobori T, Miura Y, Fukuzawa E, Yamada T, Arita T, Takenaka Y, et al. Development
and application of hysteresis steel dampers. Balkema, Rotterdam: Earthquake
Engineering, Tenth Conference; 1992. p. 1–6.
[10] Hanson RD, Xia C, SY. Design supplemental steel damping devices for buildings,
Balkema, Rotterdam: Earthquake Engineering, Tenth Conference; 1992.
[11] Kobori T. Honeycomb Damper (HDS) 2017. http://www.kobori-takken.co.jp/
English/randd/ctrl41.html.
[12] Kurokawa Y, Sakamoto M, Yamada T, Kurino H, Kunisue A. Seismic design of a tall
building with energy dissipation damper for the attenuation of torsional vibration.
Struct Des Tall Build 1998:21–32. https://doi.org/10.1002/(SICI)1099-
1794(199803)7:1<21::AID-TAL105>3.0.CO;2-8.
[13] Lynch JP. Active Structural Control Research at Kajima Corporation 1998:1–53.
[14] Onishi K, Fukuzawa E, Fukumoto T. Structural design and construction work of
Kamogawa grand tower using steel damper. Concr J 1993;31:28–38. https://doi.
org/10.3151/coj1975.31.10_28.
[15] Kareem A, Kijewski T, Tamura Y. Mitigation of motions of tall buildings with
specific examples of recent applications. Wind Struct An Int J 1999.
[16] Luth GP, Sargunuraj S, Krawinkler H, McDonald B. USC school of cinema an ex-
ample of reparable performance based design, Menlo Park, California: SEAOC 2008
CONVENTION PROCEEDINGS; 2008, p. 1–20.
[17] Li HN, Li G. Earthquake-resistant design of rc frame with “dual functions” metallic
dampers. Am Soc Mech Eng Press Vessel Pip Div PVP 2008. https://doi.org/10.
1115/PVP2007-26450.
[18] Tagami J, et al. Low cycle fatigue test of steel plate dampers with honeycomb
shaped openings. Chugoku, Japan: Summ. Tech. Pap. Annu. Meet. AIJ; 1999. p.
791–6.
[19] El-Bahey S, Bruneau M. Structural fuse concept for bridges. 9th US Natl. 10th Can.
Conf. Earthq. Eng. 2010, Incl. Pap. from 4th Int. Tsunami Symp., 2010.
[20] El-Bahey S, Bruneau M. Bridge piers with structural fuses and bi-steel columns. II:
analytical investigation. J Bridg Eng 2012;17:36–46. https://doi.org/10.1061/
(ASCE)BE.1943-5592.0000236.
[21] El-Bahey S, Bruneau M. Bridge piers with structural fuses and bi-steel columns. I:
experimental testing. J Bridg Eng 2012;17:25–36. https://doi.org/10.1061/(ASCE)
BE.1943-5592.0000234.
[22] Lee CH, Ju YK, Min JK, Lho SH, Kim SD. Non-uniform steel strip dampers subjected
to cyclic loadings. Eng Struct 2015;99:192–204. https://doi.org/10.1016/j.
engstruct.2015.04.052.
[23] Lee CH, Lho SH, Kim DH, Oh J, Ju YK. Hourglass-shaped strip damper subjected to
monotonic and cyclic loadings. Eng Struct 2016;119:122–34. https://doi.org/10.
1016/j.engstruct.2016.04.019.
[24] Farzampour A, Eatherton M. Parametric study on butterfly- shaped shear links with
various geometries. 11th US Natl Conf Earthq Eng 2018;24060.
[25] Koppal M, Eatherton MR. Perforated steel plate shear walls for tunable seismic
resistance. Struct. Congr. 2013 Bridg. Your Passion with Your Prof. - Proc. 2013
Struct. Congr., 2013.
[26] Farzampour A, Eatherton MR. Yielding and lateral torsional buckling limit states for
butterfly-shaped shear links. Eng Struct 2019;180:442–51. https://doi.org/10.
1016/j.engstruct.2018.10.040.
[27] Teruna DR, Majid TA, Budiono B. Experimental study of hysteretic steel damper for
energy dissipation capacity. Adv Civ Eng 2015:1–12. https://doi.org/10.1155/
2015/631726.
[28] Farzampour A. Evaluating shear links for use in seismic structural fuses. Virginia
A. Siar Mahmood Shah and S. Moradi
Polytechnic Institute and State University 2018. https://doi.org/10.1017/
CBO9781107415324.004.
[29] ANSYS. ANSYS Mechanical APDL 2019.
[30] Budaházy V, Dunai L. Parameter-refreshed Chaboche model for mild steel cyclic
plasticity behaviour. Period Polytech Civ Eng 2013;57:139–55. https://doi.org/10.
3311/PPci. 7170.
[31] Chan RWK, Albermani F. Experimental study of steel slit damper for passive energy
dissipation. Eng Struct 2008;30:1058–66. https://doi.org/10.1016/j.engstruct.
811
Structures 23 (2020) 799–811
2007.07.005.
[32] Farzampour A, Eatherton M. Investigating limit states for butterfly-shaped and
straight shear links, 16ECEE. USA: 16th European Conference on Earthquake
Engineering. 2018. p. 1–12.
[33] DX10.0.7. Design-Expert® Software Version 9 2017.
[34] Montgomery D. Design and analysis of experiments. Donald: Eigth. Fowley; 2013.
[35] Li HN, Li G. Experimental study of structure with “dual function” metallic dampers.
Eng Struct 2007;29:1917–28. https://doi.org/10.1016/j.engstruct.2006.10.007.